An Efficient Strategy for Adding Bulky Data into B+-Tree Indices in Information Retrieval Systems

摘要

In Internet-based information retrieval systems, a large amount of data are periodically collected from the Internet by robot agents and stored into a database. For fast retrieval, their key values are also added into an existing index, widely constructed as an inverted file using a B+-tree. Obviously, the simplest method of adding multiple key values into a B+-tree is to apply the insertion algorithm repeatedly. However, this method inserts new key values into a B+-tree in a random order without considering adjacency between them. This makes each page within the B+-tree accessed a lot of times, and thus requires large processing time. To solve this problem, we present a new method called Bulk_Add that effectively adds a bulky set of new key values into an existing B+-tree index.